As a professional sound designer you will eventually stumble across technical terms referring to audio levels and loudness. A customer might tell you: “we need a track with an integrated loudness of -23 LUFS, a dynamic range of around 15 LU and a true peak of max -1 dBTP”. Even if you’ve been producing audio for a few years you might be thinking: “huuuuuh…?”.
By now you’ve probably heard about the loudness war, and we all know what it sounds like when audio “peaks”, “clips” or hits the “red zone”. So far so good. But in order to make your audio sound pleasant and meet industry requirements, you need to know a thing or two about setting your audio levels right, and how to approach loudness when creating audio. Here’s a video of Bob Katz explaining what the loudness war does to audio in practice:
Audio produced for a specific medium should always sound pleasant, dynamic and more or less “equal” in terms of perceived loudness. Unfortunately your basic DAW meters aren’t giving you the whole truth about your audio. These meters display audio in dBFS (decibels relative to full scale) and only tell you whether your audio signal is clipping or not. To judge actual loudness, our ears respond to average levels, not peak levels.
The new standard for measuring audio loudness is LUFS (loudness units relative to full scale). It was developed to enable normalisation of audio levels, and matches how our ears actually perceive sound. LUFS meters will also tell you the integrated loudness of your audio, which is similar to RMS (root mean square) but more truthful in terms of our hearing. RMS tells you the the average power of the signal, but LUFS integrated loudness is more accurate in terms of our perception of loudness.
The integrated loudness can be understood as the “overall” level of your audio. It’s also good to keep in mind that different meters serve different purposes. You will still need to use your dBFS meters to make sure your signals aren’t clipping.
When mixing to LUFS values, your mixes will sound more consistent. The dBFS scale only measures the electrical level of the sound. If you try mixing to a level of e.g. -15dBFS, your results will likely hit different LUFS levels every time and might end up sounding different.
If you don’t yet own a LUFS meter, remember to pay attention to your RMS readings. If your dBFS peak levels go higher than -3dB, you might be in trouble. As a rule of thumb, RMS readings around -18dBFS and -14dBFS should be OK.
The new OP 59 requirements that come into effect on the 1st of January 2013 in Australia and New Zealand aim to move Australia into line with the the US, Europe and many other places in the world who now measure soundtracks based on the average perceived loudness of the soundtrack.
But how does a machine or computer measure how a human perceives loudness? I hear you ask. The answer is an algorithm developed by The International Telecommunication Union or ITU which is referred to as ITU-R BS.1770, the current evolution of which is ITU-R BS.1770-3. Using a meter that conforms to ITU-R BS.1770-3 you can actually get a quantitative measure of the loudness of a soundtrack. OP 59, like the US and European standards, specifically looks at (or listens to) the average or integrated loudness of the entire program . So if you were mixing a 30 second TV soundtrack you have to play the entire 30 seconds through the meter to arrive at your LKFS value, which for OP 59 -24LKFS. L stands for for loudness, K describes the filtering algorithm used, and FS stands for for full scale. If you are mixing a long form soundtrack for TV that was 30 minutes long you would also have to follow the same procedure although file based processing systems now can scan an audio file faster than real-time are also to be used.
The other part to OP 59 is the True Peak limit of -2dBFS. Many software limiters only measure the peak of the audio samples, not the waveform that will be created when those samples are converted back into the analog domain. This means that your audio signal may actually be distorting and clipping on playback or when being converted to lossy codecs such as MP3 or AAC which don’t handle this clipping very well and introduce distortion. Meters that conform to ITU-R BS.1770-3 should also have a true peak indication as well.
OP 59 is definitely a step in the right direction and will hopefully address the issue of jumps in loudness between ads, station promos and tv programs. Because we are now mixing to an average loudness it also means that sound engineers can be more dynamic when creating soundtracks. Basically if you add a loud bit you just need a soft bit and you’ll be back in spec (although not too soft as 1770-2 introduced a gate that removes low level audio from the calculation).
As ITU-R BS.1770-3 is an open standard many companies are now creating desktop and web based audio scanning systems which can check if audio conforms to the new loudness level requirements. Some even adjust the soundtrack so that it will meet the spec. While checking a soundtrack via this software is great from a quality control angle I think using it to normalise, or fix as some claim, the entire soundtrack is not a good idea. For instance if a soundtrack has been mixed to the old OP 48 or similar and has been heavily compressed, running it through the metering algorithm and simply normalising it to bring the loudness down will result in a mix that will sound soft and a lot less dynamic/interesting on air. For me the responsibility still lies with mix engineer to use the new freedoms/restrictions of OP 59 to their advantage and to basically know what will sound good on air and what won’t.